• 7 Posts
  • 574 Comments
Joined 2 years ago
cake
Cake day: December 1st, 2023

help-circle











  • Japanese conservative monarchists are wild.

    Look up the Google Maps reviews of the imperial palace. For some context, the majority of the imperial palace is completely off limits to the general public (in stark contrast to most developed countries), and the royal family does a new years greeting.

    The reviews are monarchists unironically saying things like that they travelled for days, lined up for hours, caught a glimpse of one of the royal family, were temporarily transported to heaven, and will dedicate their lives hoping for the forever prosperity of the royal family.





  • Unfortunately I find even prompts like this insufficient for accuracy, because even when directly you directly ask them for information directly supported by sources, they are still prone to hallucination. The use of super blunt language as a result of the prompt may even further lull you into a false sense of security.

    Instead, I always ask the LLM to provide a confidence score appended to all responses. Something like

    For all responses, append a confidence score in percentages to denote the accuracy of the information, e.g. (CS: 80%). It is OK to be uncertain, but only if this is due to lack of and/or conflicting sources. It is UNACCEPTABLE to provide responses that are incorrect, or do not convey the uncertainty of the response.

    Even then, due to how LLM training works, the LLM is still prone to just hallucinating the CS score. Still, it is a bit better than nothing.